The Power of One Sentiment Word

R Explaination

With text from the 5 Fraud articles, the subject being fraud, is likely to have many observations of the word fraud. This is a simple sentiment will show how the power of one word, in this case fraud, can seriously affect a sentiment analysis results.

Top 25 Word by Frequency, including Fraud

Top 25 Word by Frequency, not including Fraud

Conclusion

This is an extreme case, meaning fraud is the subject of all five articles from which the test was used, therefore it is bound to appear in the text at a high frequency. These slides are rather designed to demonstrate the power of one sentiment word in a sentiment analysis. Think of other documents, financial statements, xml data, html pages, etc. These all contain sentiment words:

1. Financial Documents: Asset, Assets, Liability, Liabilities, Equity, etc.
2. XML Data: Any sentiment word used in the schema or data of the xml document
3. HTML Pages: Top, Solid, Strong, etc.

Overall, the concept of this information is to ensure when you are conducting sentiment analysis to create a dictionary of sentiment words within you text which will significantly skew the sentiment of your document.

Sources

Text Mining with R, A Tidy Approach, Julia Silge & David Robinson

Articles about Fraud from Blackboard